AI Enhances Customer Service in Several Ways

SELF-SERVICE

Other companies refer to the agent augmentation as agentic AI, an advanced form of AI that possesses a degree of autonomy to solve complex, multi-step problems, Richards said. This “second phase” of GenAI, as Richards termed it, is used in self-service as well.

Unlike more traditional AI models—which rely on predefined rules and data—agentic AI can independently analyze challenges and then initiate actions and make decisions.

Beyond helping agents interact with customers, the technology is helping those customers who want to use self-service solutions, Richards said, noting, “Generative AI is being used in virtual assistants to automate the output as well as in creating conversations.”

Agentic AI applies to discrete “automations” that enable a system or program to autonomously perform a task on behalf of a user or another system (a process initiated in the background based on the situation or criteria). Its role is to act on behalf of a business (or an agent, in the case of a contact center), streamlining processes for real-time situations based on preset goals and data.

“The same AI capabilities that allow us to create a copilot solution can also be used to enhance conversational AI or chatbots,” DiAndrea added. “This significantly reduces our time to deployment of chatbots. It’s reducing the reliance on human agents for our companies and our client companies, while also reducing wait times for customers. It’s particularly valuable in contact centers where you have high volumes of customer inquiries. These AI-powered chatbots provide instant responses and are available 24 hours a day, 7 days a week.”

One example cited by DiAndrea is Maps Credit Union, which is handling 4,000 interactions a month through the NICE self-service automation solution.

INCREASING SUPERVISOR EFFECTIVENESS

The technology has increased the effectiveness and efficiency of supervisors, commented DiAndrea. “It has significantly increased the volume and quality of agent coaching, which is how we really want to see our supervisors spending their time. They don’t have to spend their time assessing calls; they don’t need to spend their time looking at alerts that might be related to average handle time or resolution. We are providing them with more contextual alerts so they can make a more data-driven decision about where they need to jump into an interaction and assist an agent. We can cut through the noise for supervisors and be more targeted in the things we ask them to do and the interventions we alert them to.”

Using GenAI in this way enabled Republic Services to increase the number of coached interactions by 120% in just 3 months, DiAndrea continued.

CODE GENERATION

While at first glance, code generation might seem to be outside of the realm of customer service, the ability to quickly generate code means better analytics to produce better customer service, DiAndrea said. “We generate a lot of data, especially when we’re thinking about solutions like interaction analytics. That data needs to be analyzed so that leaders can make data-driven decisions.”

The NICE actions solution enables managers to interact conversationally with the data so that they don’t need to write any code to develop queries or to analyze the information to determine ways to enhance customer service, DiAndrea explained.

CHALLENGES AND SOLUTIONS

Developing the above-mentioned capabilities with GenAI hasn’t come without challenges, experts agree. “The largest challenge is trust—getting companies to trust that our GenAI solutions won’t give away cars, give away airline tickets, and won’t harm the brand reputation that they’ve worked so hard to build,” DiAndrea said. “We needed to educate [prospects and customers] about the differences between some of the generic generative AI that is out there and trusted, purpose-built AI that is very specific. It’s business- specific, specific to their needs and their content, with the proper guardrails in place.”

eGain sought to build trust with its customers and prospects by starting slowly, piloting GenAI capabilities for a limited set of users, then enabling the technology for a limited set of departments, customer query types, and agent profiles (for example, experienced agents), Thalange said. “Best-practice piloting, guided by experts such as eGain’s Innovation in 30 Days no-risk, no-cost pilot, proves out GenAI and increases trust and adoption.”

Well-publicized problems experienced when using GenAI form another trust issue that customer service providers need to overcome, Thalange added. eGain’s AI Knowledge Hub uses foundational models with controls on model bias to ensure that inputs to GenAI are trusted data, context, and content to generate trusted answers. Prompts can be finetuned to reduce the chance of hallucination.

Customer service GenAI users need to be careful about using trusted data sources and monitor the output of their systems to guard against hallucinations (made-up answers), experts agreed.

“One of the biggest challenges was recognizing that guardrails absolutely have to be in place, particularly when you’re using the generative components of GenAI,” Richards said. “A lot of GenAI use RAG [retrieval-augmented generation], which means that you are using large language models [LLMs] for the generative aspects, but not for the data aspects. One of the largest challenges that organizations are overcoming now is to use their existing knowledge and existing data to augment the GenAI responses.”

Salesforce describes RAG as an AI technique that allows companies to automatically embed their most current and relevant proprietary data directly into their LLM prompt, comprising all available data, including unstructured data: emails, PDFs, chat logs, social media posts, and other types of information that could lead to a better AI output.

Security and compliance are challenges in healthcare, financial services, and any other industry or use of sensitive information, Thalange said. To overcome the risk, customer service users of GenAI need to ensure customers in their contracts that the foundational model provider doesn’t retain data or train on proprietary data, and it needs to mask any personal data.

“GenAI is not a panacea,” Richards said. “It’s a flavor of AI, but GenAI is not the appropriate AI to solve every possible problem. It’s good at generating what comes next. It’s good at recognizing patterns and capitalizing on that. It’s not a substitute for machine learning, though. It’s not going to do everything.” The technology will not automate all customer interactions, she added.

LOOKING AHEAD

By the end of the year, DiAndrea expects to see deepening and further improvements in the GenAI use cases.

“Those solutions are already delivering business value. As we build our road map for each of these capabilities, we’re seeing new and emerging use cases within each. Over the next year, I think we’re going to see GenAI further tackle challenges like reducing service response times, becoming even more personalized at scale across trillions of transactions, and proactively identifying and addressing customer needs before they are ever voiced.”

Brands have cited advanced sentiment and intent analysis of customers’ queries as a top priority and trend for next year, according to Dvir Hoffman, CommBox CEO. “By integrating customer sentiment data with other behavioral metrics (such as purchasing history and engagement levels), brands can gain a richer, more holistic understanding of their customers, leading to smarter strategies for retention and satisfaction.”

For example, if a customer shows frustration or dissatisfaction, your service team could be alerted to prioritize the issue and respond with extra care, Hoffman explained. “Coupled with advanced automation capabilities, this level of personalization and service doesn’t always need to come from an agent—you can deliver this experience from your chatbot—an AI bot that can use customer sentiment and respond promptly or route the conversation to a specialist human agent if needed.”

Thalange expects to see the emergence of dynamic AI orchestration, with the collaboration of GenAI, AI reasoning, bots, and business logic with human experts in the loop, to enable assured conversations and complex problem solving.

“First movers will reap sustainable competitive benefits by outgrowing slower competitors,” Thalange added. “GenAI will go from pilot to a transformative enabler across business functions and industries, resulting in many success stories.”